- A human talent shortage handicaps many companies that are getting started with AI
- Colleges and universities are starting to fill AI skill gaps in the marketplace
- Teaching existing employees AI skills can be an equally viable strategy
As companies scramble to figure out how artificial intelligence can give them a competitive edge, many are tripped up by a critical prerequisite: finding enough skilled AI professionals.
In fact, more than half of senior AI professionals believe a talent shortage is the biggest obstacle to implementation, according to research from Ernst & Young.
“The talent war for these professionals is very real,” says Lane Greever, a senior VP at IT staffing company Modis. “There are so few people with degrees in this area or ones who have done this work for a long time that they’re in incredibly high demand.”
One estimate pegs the number computer scientists capable of building AI systems at 22,000. Among these highly skilled individuals, only about 3,000 are looking for a job at any given time. Meanwhile, top companies have about 10,000 open positions for these people.
How can companies close the skills gaps in AI? By thinking outside the current labor market.
With today’s demand far outpacing supply, companies are turning to colleges and universities to snap up talented, greenhorn graduates, while others look to online trainings programs to up‑skill or re‑skill employees they already have.
Building the undergrad pipeline
Many colleges have begun adding undergraduate AI and machine learning curricula, ranging from concentrations in computer science to degrees backed by research institutes.
The University of Massachusetts at Amherst, for example, offers a B.S. in computer science with an AI degree track focused on the fundamentals of constructing intelligent, autonomous systems. Students at the University of Illinois can pursue a B.S. in computer engineering with a specialization in AI, robotics and cybernetics, the science of communication and control systems.
University of Massachusetts undergrads take courses in software engineering, algorithms, natural language processing, machine learning, and robotics—all disciplines in big demand, says James Allan, professor and chair at the university’s College of Information and Computer Science.
“Just being exposed to machine learning or the tools used in the workplace makes these students valuable,” he says. “Many go into software engineer roles and are taking jobs in industries that five years ago you’d never think had anything to do with data science or machine learning.”
Many graduates emerge with marketable skills such as neural network programming, says Mark Hasegawa‑Johnson, professor of electrical and computer engineering at the University of Illinois. His students have recently taken jobs working on self‑driving cars, automatic speech recognition, facial‑recognition technologies, automated investing and disease diagnosis.
“There’s a lot of flexibility with graduates, and businesses like them because there’s a level of certification with the skills they have—but they’re getting snapped up quickly, often after completing an internship.” One of Allan’s Ph.D. students drew a job offer more than a year before she completed her degree.
One challenge for businesses in recruiting graduates—aside from funding salaries that can reach $300,000 for newly minted Ph.D.s—is applicable experience and depth of knowledge, Greever says. “We’re seeing that businesses are having to relax their hiring requirements because they can’t always find what they’re looking for in a college graduate.”
Companies hiring graduates with basic AI and machine learning skills often start them out in a training program. Other businesses have adopted a model in which fresh grads spend 20 percent of their time working on projects to gain more experience.
“Because there’s such a huge demand for people in AI and ML and other related fields—and such a scarcity of people graduating with this background—companies are starting to look a little more carefully about what they actually want,” Allan says.
Because candidates with AI skills and experience are so scarce, some businesses opt to retrain people they already have. “These businesses might have employees with a penchant for coding, data, and analytics, and so they’re reaching out to training companies and immersion programs to get their people up to date on the latest skills,” Greever says.
Udacity, an online learning platform that uses video instruction, interactive content, mentors, and reviewers, offers what it calls Nanodegrees—project‑based online certifications on a variety of topics that can be completed in three to 12 months, and are available for both businesses and individuals.
The company’s deep learning and AI programs are among its most popular offerings, says Mat Leonard, head of AI learning at Udacity.
“The last decade or so was all about mobile, and now everyone is focusing on machine learning,” Leonard says. “The popularity of our AI courses reflects what’s happening inside businesses. They’re hopping on the wagon, and we’re working to meet the demand.”
Udacity courses are taught by a variety of people from academia and industry. The AI nanodegree program includes instruction by Udacity co‑founder Sebastian Thrun, who also founded Google’s self‑driving car team. About half the students who graduate from the program have found jobs or received promotions after completing it, Leonard says.
These training options are attractive to businesses. By offering training within the organization, companies are able to bypass timely and costly talent searches while unifying the skills employees learn through one consistent platform.
“Companies need to build a pipeline for the future and build these skillsets,” says Allan Leinwand, ServiceNow’s chief technology officer. “By looking at all the options, whether it’s retraining employees, hiring graduates or smarter onboarding, companies can augment new technologies to enhance the world of work.”